An AI industry of more than $1 trillion that can’t ship any functional production code Umkhakha we-AI we-$1 trillion eyenza ukuba awukwazi ukuhanjiswa nayiphi na ikhowudi yokukhiqiza Ukuba unemibuzo malunga ne-apocalypse ye-AI eyenziwe ngalinye, nceda ukhangele: emva kokusetyenziswa kwezigidi ezigidi, sinokufumaneka ukunikezela ikhowudi olufanelekileyo kunye nezilwanyana ze-AI ezidlulileyo. (ngoku) (Nceda nqakraza ngakumbi kwi-blast ye-AI bubble) Qhagamshelane iingcebiso ze-"Breakthrough". Qhagamshelane iingcebiso ze-CEO. Ngoku qhagamshelane i-catalgia: i-exponential surge of garbage code, ezisetyenziselwa i-CEO (Khona, akukho "izibonelelo", nto leyo yokuthengisa, akukho nenzululwazi) ukuba akukho umntu wamkelekile. same four errors Uyaona abo ezine kwifowuni elandelayo. Emva koko siza kuhamba ngeengxelo zayo zayo zibonakalayo, ukuze unokufumana ngexesha elifanelekileyo kwi-AI-generated code yakho. Yintoni zibonakalayo ngokwenene? Izindlu ze-OpenAI Iingxaki ezine ezine. I-Google Rolls Out Iingxaki ezine ezine. I-Anthropic Unveils Iingxaki ezine ezine. Microsoft iimveliso yonke into - Akukho izihlangu apho, ke? GPT-5 Gemini 2.5 Claude Sonnet 4.5 Copilot 
 
 Kwiminyaka emininzi, iingxowa zihlanganisa: i-spaghetti ye-code, i-security holes ezitsha kunye ne-punchline efanayo - i-AI code ibonisa smart, ibhekwa ngempumelelo, kwaye nangokufuneka abantu ukucacisa i-confusion. Kwiminyaka emininzi, iingxowa zihlanganisa: i-spaghetti ye-code, i-security holes ezitsha kunye ne-punchline efanayo - i-AI code ibonisa smart, ibhekwa ngempumelelo, kwaye nangokufuneka abantu ukucacisa i-confusion. Iimpawu ziquka ngokucacileyo 
 
 
 
 
 I-Churn ye-Short-Term iye yandisa i-100% ukususela kwi-2021. Izixhobo ezisetyenziswa kwe-AI zibonisa i-40% ezininzi iingxaki zokhuseleko I-developer eyenziwe ngexesha le-19% ngexabiso nge-AI tools Iingcali kwiinkampani ezininzi ezahlukileyo zihlanganisa malunga ne-70% yeengxaki ze-AI. Rose ngeenxa 100% Rose ngeenxa 100% I-40% ezininzi iinkcukacha zokhuseleko I-40% ezininzi iinkcukacha zokhuseleko 19% engaphantsi 19% engaphantsi Izixhobo ze-AI Ukukhusela malunga 70% Ukukhusela malunga 70% Iinkampani ezininzi zilungiselela yonke into kwi- "i-10x productivity", i-impossibility ye-mathematic eyenza ngoku ekwenzeni isixeko ngaphandle kwimveliso ye-software. I-We are not building the future faster, i-We are building Kwi-scale engabonakaliweyo, kunye ne-confidence engabonakaliweyo, kunye ne-blindness engabonakaliweyo kwi-catastrophe eyenza. Here’s what’s really happening: tomorrow’s minus two trillion dollars of code Nceda nceda nceda ukubonisa indlela ezininzi Zibonisa i-codebase yehlabathi kwi-catastrophe engabikhoyo. Ukuba unomatshini we-software, uqhagamshelane, ngokukhawuleza ibhokisi lakho lokufumana i-imeyile ze-e-mail ezincinane ezivela kumazwe ngamazwe omnqweno omnye yokukhusela i-codebase yayo. Four Horsemen of Bad Code Ndiya, uya kufumana ngoko ke iinkampani ziya kuba - emva kokufumana i-batch ye-The iimvakalelo desperately rehiring the same human programmers they replaced “we won’t need programmers in a few months” "Kuba iinyanga eziliqela, siya kufuneka i-programmers" I-Four Horsemen Riding On Tons Of Garbage Code Njengoko kulandwa, iingxaki ezine ezibonakalayo kwi-diagram phezulu ziquka i-AI-generated code eningi. Iinkampani zihlanganisa ngesantya epheleleyo, kodwa i-catastrophe kunokutshiswa kuphela. Kwangaphambili okanye elandelayo, kuya kufuneka ukholelwa ukuba ingxaki ye-AI kufuneka isetyenziswe ukuba yintoni: Ngaphandle kwe-automation autocomplete on steroids Ukunciphisa ukubaluleka kwamanzi ezininzi iimpawu ezidlulileyo, kufuneka siqhelise kwiindidi ezimbini. Iintlobo ezimbini zokuqala ziquka wonke umntu uyazi; iiyure ezidlulileyo , ebizwa kwi-mechanics ye-AI ngokufanayo, ezininzi abafazi. Zonke iinkcukacha zithunyelwe ngokufanelekileyo njengoko , i-buzzword efunyenwe ngenxa yentengiso, akukwazi ukuqeqesha. classic statistical errors new cognitive errors “I-Hallucinations” Iingxaki ezimbini eziqhelekileyo bonke bayazi Type II Errors : The “Obvious” Mistakes Iingxaki ezincinane ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye zibe. What it means (in code): # AI generates (wrong parameter use; forgot to apply discount)
def calculate_discount(price, customer_id, discount_rate):
    return price * discount_rate  # Oops—no discount applied

# Should have been
def calculate_discount(price, customer_id, discount_rate):
    return price * (1 - discount_rate)  # Apply the discount properly
 Izibonelo ezininzi ezidlulileyo kwi-AI-generated code zihlanganisa: 
 
 
 
 
 
 Ukusetyenziswa kwamafutha (i-Try / Catch Blocks) I-Operators ezifanelekileyo (i-using = instead of == in conditions) Iimpawu ze-off-by-one kwi-loops Iimpawu ze-method (string.contains() kwi-Python kunokuba kwi-operator) iiyure elide etc. Type II Errors (Logic/Causal): Wrong Business Logic, Perfect Syntax Ukusebenza ngokugqithisileyo; uyenza into olungabikhoyo. Okanye, ukuba banxibele kakhulu, uyenza Ukucinga What it means (in code): right on time… at the wrong station. Case-in-point - kwaye unako ukuvelisa oku kwi-business yakho: i-AI wabhala ukuba i-80% yabasebenzisi ze-premium kwiinkcukacha zayo zonyango kwiidolophu ezinkulu, ngoko ilungele ngempumelelo ukuba Ukucinga . city location shipping rates # AI generates (correlation != causation)
def determine_shipping_cost(user):
    # AI saw premium users often live in cities with free shipping
    if user.is_premium:
        return 0  # Free shipping for premium
    elif user.city in ['New York', 'Los Angeles']:
        return 0  # AI incorrectly learned cities get free shipping
    else:
        return 9.99

# Should have been based on actual business rules
def determine_shipping_cost(user, order_total):
    if order_total > 50 or user.is_premium:
        return 0
    return 9.99
 I-New Two Cognitive Errors I-Nobody Talks About Iingxaki ezidlulileyo ezimbini zibonakalayo kwi-cognitive. Ungayifumana kwiingxaki ze-probability ezidlulileyo. Zibonakalisa ukuba zibonakalayo kwiinkqubo ze-AI, ezivela ukususela njani iimodeli ezintsha zihlanganisa iingcebiso kunye nokunxibelelana ngexesha lokufunda kunye nemveliso. Zibonakalisa ngezindlela ezahlukeneyo: enye njengoko iinkcukacha, enye njengoko Ukuthetha njani iingcamango kunye nokufunda. entanglement collisions Type III Errors (Spaghetti/Entanglement): When Everything Gets Tangled Yinto yokuqala yokuzonwabisa. Iindawo ezahlukileyo ze-systems zihambe: ukuguqula i-imeyile, kwaye iintlawulo zihlukile; ukuguqula i-auth, kunye ne-logging zihlukile. Akukho into esebenzayo ngalinye, ngoko zonke iingcebiso ezincinane zihlukile kwi-codebase epheleleyo. Yinto yinto efumaneka xa i-AI ibandakanya iingcebiso kwaye ibandakanya i-spaghetti ye-architectural. # AI generates - everything tangled together
def process_user_action(user, action_type, data):
    # User validation, logging, auth, payment, and email ALL mixed
    if not user.is_authenticated:
        log_event("auth_failed", user.id)
        send_email(user.email, "Login attempt failed")
        return False
    
    if action_type == "purchase":
        # Purchase logic mixed with user verification
        if user.email_verified and user.age > 18:
            payment = process_payment(data['amount'])
            log_event("purchase", user.id, payment.id)
            update_inventory(data['item_id'])
            send_email(user.email, "Purchase confirmed")
            update_user_points(user, data['amount'] * 10)
            check_fraud(payment)  # Why is fraud check AFTER payment?
            return payment
    
    elif action_type == "login":
        # Login mixed with analytics and recommendations
        update_last_login(user)
        log_event("login", user.id)
        recommendations = get_recommendations(user)  # Why here?
        send_email(user.email, "Welcome back!")  # Email on EVERY login?
        return {"success": True, "recommendations": recommendations}
 Summing up, you can’t: 
 
 
 
 
 Ukuguqulwa kwe-email logic ngaphandle kokuphendula iintlawulo Ukuguqulwa kwe-authentication ngaphandle kokutshintsha i-logging Ukuhlaziywa kwifowuni yokupakisha ngaphandle kokusebenzisa i-User Verification Ukubuyekeza zonke iinkcukacha kwi-isolation Yintoni iingxaki zibonisa Yintoni kufuneka ngokupheleleyo Ngenxa yokuba nje engabonakaliwe. 100% increase in code Rewritten ngaphakathi 2 iiveki Rewritten ngaphakathi 2 iiveki Type IV Errors (Memory / Collision Cascades): Unrelated behaviors share the same “address” Njengoko kulandwa, uya kufumana kwiincwadi ze-probability, kodwa, ngokwenene, ibhale kwi-AI-generated code yakho. 
 
 Inkcazo elula le mfuneko: iingubo ezimbini kwi-stick enye; omnye uyafumaneka xa unayo omnye. Inkcazo elula le mfuneko: iingubo ezimbini kwi-stick enye; omnye uyafumaneka xa unayo omnye. Ukubonisa xa izibonelelo ze-compressed zihlanganisa iimpawu ezahlukeneyo kakhulu. Kwixesha lokugqibela, ukufumana omnye ngokuhambisa omnye. Uyafuna i-password, kunye ne-account cleanup tags. Uxhumane iinkcukacha, kwaye i-payment ebandayo isebenza. Akukho into ebandayo emzimbeni, kodwa iziphumo ezininzi zithembisa ngenxa yokuba iimodeli ibekwe nabathengi kwixesha elifanayo. # AI generates — password reset collides with account cleanup
def reset_password(user_email):
    user = find_user(email=user_email)
    if user:
        token = generate_reset_token()
        send_reset_email(user.email, token)
        
        # From a different flow, yet here we are
        if user.last_login < 30_days_ago:
            user.status = "inactive"
            cleanup_user_data(user)
        
        return token
    return None

# Another function unexpectedly handles payments
def export_user_data(user_id):
    data = get_user_data(user_id)
    
    # Payment processing in a data export—what could go wrong?
    if data.outstanding_balance > 0:
        process_payment(data.outstanding_balance)
    
    return create_export_file(data)
 Real-world echoes (reported in research): 
 
 
 
 
 Iimpawu ze-Search ezintsha iidipozithi ze-DB I-Auth Flows Eyenza I-Analytics Ukubonisa i-endpoints ezivela kwi-logs I-APIs ezihlanganisa iintlobo ezininzi ezahlukileyo Indawo yokuguqula lula: ukunika iimeko namespaces ezahlukileyo kunye neengxaki ezizodwa, ngoku uqhagamshelane into eyahlukileyo nge-path dedicated that you can test on its own. Use retrieval grounding, key-value memory, and "fail-closed" side effects to keep things from colliding. 
 
 Ndiya, ndiyazi, kwaye ndiyazi ukutya ukuba bafumene into efanelekileyo kwaye ngakumbi. Ndiya, ndiyazi, kwaye ndiyazi ukutya ukuba bafumene into efanelekileyo kwaye ngakumbi. Yintoni iingxaki zikhusela software yokukhiqiza? Ngiyazi le nqaku wonke umdla – into efanelekileyo kwi-coding AI ships. I-Chart 3 apha ibonisa i-bugs ezininzi eziqhelekileyo abaphelisiwe yi-developer ehlabathini lonke kunye neendlela eziqhelekileyo kwiintlobo ezine ze-error (iintlobo ezintathu ezininzi ezininzi kwi-practice). I-Chart 2 apha kunye ne-Chart 4 apha ibonisa into efanelekileyo xa abathengi bafumene i-AI assistants kwi-workflow. Xa isetyenziswe ngqo njengoko , I-Type I-II amaxabiso ezininzi zithathwa ngabantu okanye izixhobo. Uyakwazi ukufumana i-productivity bump - i-typos i-squashed, i-import fixed, i-homework ebonakalayo. Nice, kodwa engabonakali i-revolution. autocomplete on steroids Yiba — i-spaghettization kunye neengxaki ze-semantic. Yinto xa amaqhosha zithunyelwe. Iingcebiso zithunyelwe kwiimoduli, iingcebiso zithunyelwe, kwaye abathengisi "ukukhuselekileyo" ziye ziye ziye ziye ziye zithunyelwe ngempumelelo. Umphumela? I-code ebandayo ebandayo, i-testing, kunye ne-inflates izindleko zokusebenza. Iingqungquthela ezininzi zithunyelwe ngempumelelo ngokukhawuleza - kunye ne-ukhuseleko - ukuchithisa iindawo ezisekelwe yi-AI kunokuba zithunyelwe ngexesha elide. Types III–IV Kwaye ngoko ke, unokufumana ngokucacileyo ukuba i-dream ye-"10x automation" iyafumaneka, i-ROI iyafumaneka ngokupheleleyo kwaye ibonisa ukuba, ukususela kwi-2025, i-code-generation automation yayo, kwaye nayiphi na ukuveliswa kwe-productivity ye-developer emzimbeni, akuyona kuphela i-phantasy elide, epic. Ngoko ke, ukucacisa: njenge-developer ye-software, unemfuneko ukuxhaswa ngexesha elinye i-doom-poster iye i-snake-oil entsha "i-AI code wand." Njengoko i-Chart 4 apha ibonisa, i-ROI yokufuneka kuphela, kwaye ibonisa kuphela xa usebenzisa i-AI kwiingxaki ezincinane kunye ne-low-risk izinto ezilungileyo, i-lint, kunye ne-roll-back. Ukuba a . power-autocomplete robot engineer 
 
 
 
 
 
 I-Boilerplate / i-staffolding: i-CRUD handlers, i-DTOs, i-wire Ukucaciswa kwe-lim: i-stubs, i-test ye-table-driven, i-fixture data Ukucaciswa kwezobugcisa: ama-refactors ezininzi; i-rename / i-move ops Iinkcukacha ezincinane: i-docs/comments, i-regexes, i-SQL elula, iimveliso ze-migration I-Housekeeping: i-lint / i-format hints; i-config & i-YAML snippets Nazi iindaba elungileyo ukuba wonke umdlali unemibuzo: yonke "10x nge-AI" fantasy kuxhaswa kwimveliso. Iinkampani ziya kufuneka Ukuqhathanisa i-spaghetti, ukuguqula i-logic, kunye nokuguqula iinkqubo ezisebenza ngokwenene. Ngoko uqhubeke ukuqhathanisa izilwanyana zakho. Iimpahla ze-cleaning ziye zihlanganisa, kwaye zibonise ukuba, yup, real programmers — lots of them — you’re on the call list. Ukucaciswa Iingcebisi zeengcebiso zethu ezidlulileyo bayazi ukuba i-antidote ikhona - kodwa ngokwenene, oku kuthetha ukunciphisa, ukuguqulwa kwinkqubo, kwaye ufumane ukuba i-hype ayikho nophando. Kukho ingozi, ngaphandle kokufuneka, ukuba umntu uye uqhuba i-brakes kunye nokufunda i-science engaphezulu kunye ne-marketing engaphansi. Ukusetyenziswa kwe-matematical ye-four-error plague Ukusetyenziswa kwe-matematical ye-four-error plague

Read My Stories

Le audio iveliswe ngolwimi lokuqala lwebali!

Inde kakhulu; Ukufunda

Iinkampani ziye ngokushesha ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye zibe.

Iinkampani ziye ngokushesha ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye zibe.

About Author

IZIMVO

ZIJONGE IIMPAWU

ELI NQAKU LINIKEZELWE KU

Related Stories

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps